Content
85%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is an unusually strong operational skill: fully executable commands, explicit dry-run and verification checkpoints, feedback loops, and a precise hard-stops section. Its weaknesses are moderate verbosity in the Plite Release Claims section and a monolithic single-file structure where a 220-line skill with a detailed script-behavior contract would benefit from one-level-deep reference files.
Suggestions
Split the sync-script behavior contract ("The script must:" bullet list) and the verify command sets into a reference file (e.g. references/sync-contract.md), keeping SKILL.md as the overview with clearly signaled links.
Tighten the Plite Release Claims section to a single sentence of requirements plus one sentence of prohibition, removing the repeated framing.
Trim meta-commentary in "Core Take" (e.g. "This fast mode is the whole direct sync recipe") to pure instruction.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with operational specifics (exact gh/git/npm commands, guards, hard stops) and explains nothing Claude already knows. Scored 4 not 5 because the "Plite Release Claims" section repeats its caution across multiple sentences and "Core Take" carries meta-commentary that could be trimmed; not 3 because the excess is minor, not systemic. | 4 / 5 |
Actionability | Fully executable, copy-paste-ready commands throughout: "git worktree add --detach ../plate-sync origin/next", "gh workflow run promote.yml --ref next -f dry_run=true", "PSTACK_RELEASE=1 git push origin next", plus exact JSON field lists for gh run list. The remaining placeholders (<release-run-id>) are appropriately contextual. | 5 / 5 |
Workflow Clarity | Clear sequencing via the mode table, dry-run-before-execute checkpoints in both sync and promote flows, "gh run watch --exit-status" feedback loops, post-publish dist-tag verification, an explicit Hard Stops list, and error-recovery guidance ("If the script stops on files outside known release metadata, do not guess"). Scored 5 not 4 because validation checkpoints are explicit and pervasive rather than minor. | 5 / 5 |
Progressive Disclosure | Sections are well-organized and navigable, but this ~220-line skill is entirely inline with zero reference files; content that could live in separate references (the ~20-line sync-script behavior contract, the verify command sets) is inlined in SKILL.md. Scored 3 not 4 because no content is actually split out; not 2 because structure is good and nothing is buried. | 3 / 5 |
Total | 17 / 20 Passed |